In well-funded healthcare settings, robotic surgery is gathering momentum. The technological rate-limiting steps are rapidly being tackled: progressive miniaturisation of components, enhanced optics, and progress with AI are important steps. Currently, these robots work largely as an extension of a human surgeon operator and are certainly far from autonomous. For example, the human-controlled da Vinci system (Intuitive Surgical, USA) is probably the best known.
Whilst robotic surgical systems are building a progressively solid evidence base (in the right setting), there are some important questions to be considered. Utilising robots adds complexity to an already complex arena. Just as humans can ‘break-down’, or surgical kit can malfunction, so too can mechanical or electronic systems. However, if robots can be incorporated safely to enhance performance (and therefore also improve patient care and outcomes) it is inevitable that the technology will be embraced and will develop.
There is, however, another very large elephant in the operating room.
Consider cars. They are gradually becoming more autonomous (by way of 3D cameras, AI, and constant connectivity) and will ultimately become driverless. The rail and aviation industries are heading in a similar direction. Will we ultimately be operated upon by fully autonomous surgical robots?
Robots that are able to make decisions and learn from experience are already in existence. Recent advances in machine learning are well publicised. As time passes, the control ‘dial’ is likely swing away from humans and towards more isolated, autonomous robotic systems. This redistribution of decision-making and motor performance poses a legal question. Who would be liable in the event of a mistake? Moreover, problem solving strategies utilised by machine learning algorithms are (almost by definition) not accessible to an average human mind – resulting in a ‘black box’ of decision-making that will be very hard to unpick.
What this means in practise is hard to predict. A human surgical supervisor (at the very least) is probable for some time yet. From a governance perspective, it has already become urgently necessary to establish international standards for the use of AI in health care – to ensure accountability and to protect patients. All this aside, it is worth reiterating that operating is just one part of being a surgeon. Human to human interaction is the bedrock of the therapeutic alliance needed to provide surgical care for people. Robots will struggle to fulfil this role.
Consultant Neurosurgeon and Honorary Senior Lecturer, Edinburgh